Targeting Aldehyde Dehydrogenase Cancer Stem Cells in Ovarian … · Aldehyde dehydrogenase-1A1 (ALDH1A1) expression characterizes a subpopulation of cells with tumor-initiating or

Therapeutic Discovery

Targeting Aldehyde Dehydrogenase Cancer Stem Cellsin Ovarian Cancer

Charles N. Landen Jr1, Blake Goodman2, Ashwini A. Katre1, Adam D. Steg1, Alpa M. Nick2, Rebecca L. Stone2,Lance D. Miller3, Pablo Vivas Mejia4,5, Nicolas B. Jennings2, David M. Gershenson2, Robert C. Bast Jr.4,Robert L. Coleman2,6, Gabriel Lopez-Berestein4,6,7, and Anil K. Sood2,6,7

AbstractAldehyde dehydrogenase-1A1 (ALDH1A1) expression characterizes a subpopulation of cells with

tumor-initiating or cancer stem cell properties in several malignancies. Our goal was to characterize

the phenotype of ALDH1A1-positive ovarian cancer cells and examine the biological effects of ALDH1A1

gene silencing. In our analysis of multiple ovarian cancer cell lines, we found that ALDH1A1 expression

and activity was significantly higher in taxane- and platinum-resistant cell lines. In patient samples, 72.9%

of ovarian cancers had ALDH1A1 expression in which the percentage of ALDH1A1-positive cells

correlated negatively with progression-free survival (6.05 vs. 13.81 months; P < 0.035). Subpopulations

of A2780cp20 cells with ALDH1A1 activity were isolated for orthotopic tumor–initiating studies, where

tumorigenicity was approximately 50-fold higher with ALDH1A1-positive cells. Interestingly, tumors

derived from ALDH1A1-positive cells gave rise to both ALDH1A1-positive and ALDH1A1-negative

populations, but ALDH1A1-negative cells could not generate ALDH1A1-positive cells. In an in vivo

orthotopic mouse model of ovarian cancer, ALDH1A1 silencing using nanoliposomal siRNA sensitized

both taxane- and platinum-resistant cell lines to chemotherapy, significantly reducing tumor growth in

mice compared with chemotherapy alone (a 74%–90% reduction; P < 0.015). These data show that the

ALDH1A1 subpopulation is associated with chemoresistance and outcome in ovarian cancer patients, and

targeting ALDH1A1 sensitizes resistant cells to chemotherapy. ALDH1A1-positive cells have enhanced,

but not absolute, tumorigenicity but do have differentiation capacity lacking in ALDH1A1-negative cells.

This enzyme may be important for identification and targeting of chemoresistant cell populations in

ovarian cancer. Mol Cancer Ther; 9(12); 3186–99. �2010 AACR.

Introduction

Ovarian cancer was expected to be diagnosed in 21,550women in 2009 and take the lives of 14,600 women (1).Although ovarian cancer is among the most chemosensi-tive malignancies at the time of initial treatment (surgery

and taxane/platinum-based chemotherapy), mostpatients will develop tumor recurrence and succumbto chemoresistant disease (2). An understanding of themechanisms mediating survival of subpopulations ofovarian cancer cells is necessary to significantly improveoutcomes in this disease.

In many malignancies, a subpopulation of malignantcells termed cancer stem cells or tumor-initiating cells hasbeen hypothesized to represent themost tumorigenic andtreatment-resistant cells within a heterogeneous tumormass. Defined by their enhanced ability to generatemurine xenografts and give rise to heterogeneous tumorsthat are composed of both tumor-initiating cell and non-tumor-initiating cell populations, these cells may also bemore chemoresistant and depend on unique biologicalprocesses compared with the majority of tumor cells (3,4). In ovarian cancer, many of these properties have beenidentified in populations of CD44/c-kit–positive cells (5),CD133-positive cells (6–8), and Hoechst-excluding cells(the side population; ref. 9).

Among several markers that have been used to iden-tify cancer stem cells, aldehyde dehydrogenase-1A1

Authors' Affiliations: 1Department of Obstetrics and Gynecology, Uni-versity of Alabama at Birmingham, Birmingham, Alabama; 2Department ofGynecologic Oncology, U.T.M.D. Anderson Cancer Center, Houston,Texas; 3Department of Cancer Biology, Wake Forest University Schoolof Medicine,Winston-Salem, North Carolina; 4Department of ExperimentalTherapeutics, U.T.M.D. Anderson Cancer Center, Houston, Texas; 5TheUniversity of Puerto Rico Comprehensive Cancer Center, Rio Piedras, PR;and 6Center for RNA Interference and Non-Coding RNA; 7Department ofCancer Biology, U.T.M.D. Anderson Cancer Center, Houston, Texas

Note: Supplementary material for this article is available at MolecularCancer Therapeutics Online (http://mct.aacrjournals.org/).

Corresponding Author: Charles N. Landen Jr., Department of Obstetricsand Gynecology, The University of Alabama at Birmingham, 1825 Uni-versity Blvd, 505 Shelby Bldg, Birmingham, AL 35294. Phone: 205-934-0473; Fax: 205-934-0474. E-mail: [email protected]

doi: 10.1158/1535-7163.MCT-10-0563

�2010 American Association for Cancer Research.

MolecularCancer

Therapeutics

Mol Cancer Ther; 9(12) December 20103186

on July 27, 2021. © 2010 American Association for Cancer Research. mct.aacrjournals.org Downloaded from

Published OnlineFirst October 1, 2010; DOI: 10.1158/1535-7163.MCT-10-0563

http://mct.aacrjournals.org/

(ALDH1A1) has been a valid marker among severalmalignant and nonmalignant tissues (10–20). It holdsthe attractive distinction of not only being a potentialmarker of stemness but potentially playing a role in thebiology of tumor-initiating cells as well (10). ALDH1A1,1 of 17 ALDH isoforms, is an intracellular enzyme thatoxidizes aldehydes, serving a detoxifying role, andconverts retinol to retinoic acid, mediating control ondifferentiation pathways. The ALDH1A1 populationdefines normal hematopoietic stem cells, being usedto isolate cells for stem cell transplants in patients.Using the ALDEFLUOR assay, a functional flow cyto-metric assay that identifies cells with active ALDH1A1,tumor-initiating cell-enriched populations have beenidentified in multiple malignancies (20), includingbreast (11–14), colon (15, 16), pancreas (17), lung (18),and liver (19). Whether or not the ALDH1A1-active population is enriched for tumor-initiating cellshas not been demonstrated for ovarian cancer. Moreimportantly, although ALDH1A1 is implicated inchemoresistance pathways, it is not known whethertargeting ALDH1A1 can sensitize resistant cells tochemotherapy and therefore represent a potential targetfor cancer stem cell–directed therapy. We sought tocharacterize expression of ALDH1A1 in ovarian cancercell lines and patient samples, determine whether itcontains tumor-initiating cell properties, and examinewhether targeting ALDH1A1 sensitizes cells to che-motherapy in both in vitro and in vivo ovarian cancermodels.

Materials and Methods

Cell lines and cultureThe ovarian cancer cell lines SKOV3ip1, SKOV3-

TRip2, HeyA8, HeyA8MDR, A2780ip2, A2780cp20,IGROV-AF1, and IGROV-cp20 (21, 22) were maintainedin RPMI-1640 medium supplemented with 15% fetalbovine serum (Hyclone). SKOV3TRip2 [taxane-resis-tant, a kind gift of Dr. Michael Seiden (23)] andHeyA8MDR were maintained with the addition of150 nmol/L of paclitaxel. The HIO-180 SV40-immorta-lized, nontumorigenic cell line derived from normalovarian surface epithelium was a kind gift of Dr.Andrew Godwin. All cell lines were routinely screenedfor Mycoplasma species (GenProbe detection kit) withexperiments done at 70% to 80% confluent cultures.Purity of cell lines was confirmed with STR genomicanalysis, and cells used were always less than 20 pas-sages from the stocks tested for purity.

Whole genomic analysisRNA was extracted from 3 independent collections of

SKOV3ip1 and SKOV3TRip2 cells at 80% confluencewith the RNeasy Mini kit (Qiagen). It was subjectedto microarray analysis using the Illumina HumanRef-8Expression BeadChip, which targets�24,500 well-anno-tated transcripts. Microarray data were normalized by

the cubic-spline method (24) using the Illumina Bead-Studio software. The significance of differentiallyexpressed genes was determined by Student’s t testfollowed by correction for false discovery (25). A heatmap was generated using Cluster 3.0 and Java TreeViewsoftware. The array data have been registered with GEO(accession #GSE23779) for public access.

Western blot analysisCultured cell lysates were collected in modified radio-

immunoprecipitation assay lysis buffer with proteaseinhibitor cocktail (Roche) and subjected to immunoblotanalysis by standard techniques (26) using anti-ALDH1A1 antibody (BD Biosciences) at 1:1,000 dilutionovernight at 4�C, or anti–b-actin antibody (Sigma Che-mical) at 1:2,000.

Immunohistochemical staining andclinical correlations

Immunohistochemical (IHC) analysis was done onformalin-fixed, paraffin-embedded samples, using stan-dard techniques (26). For ALDH1A1, antigen retrievalwas in citrate buffer for 45 minutes in an atmosphericpressure steamer, using anti-ALDH1A1 antibody (BDBiosciences) at 1:500 dilution in Cyto-Q reagent (InnovexBiosciences) overnight at 4�C. Primary antibody detectionwas with Mach 4 HRP polymer (Biocare Medical) for20 minutes at room temperature, followed by diamino-benzidine incubation. After IHC staining, the numberof tumor cells positive for ALDH1A1 was countedand expressed as a percentage of all tumor cells byan examiner blinded to clinical outcome. Patient sampleswere categorized as having low (<1%), intermediate(1%–20%), or high (21%–100%) ALDH1A1 expression.The IHC analysis was done on samples collected atprimary debulking surgery on 65 untreated patientswith stage III–IV, high-grade papillary serous adenocar-cinoma; with institutional review board approval, clinicalinformation was collected. Progression-free and overallsurvival were plotted with the Kaplan–Meier methodfor patients in each group of ALDH1A1 expressionand compared with the log-rank statistic by using PASW17.0.

For dual staining of ALDH1A1 and CD68 (for macro-phages), staining for ALHD1A1 was done first as pre-viously, followed by exposure to anti-CD68 antibody(1:4,000; Dako) and goat anti-mouse-AP (Jackson Immu-noresearch). AP was developed with Ferangi Blue chro-magen kit (Biocare Medical). For dual staining ofALDH1A1 and hypoxic tumor regions, mice bearingSKOV3TRip2 xenografts were injected with 60 mg/kgof Hypoxyprobe-1 reagent (HPI, Inc.). Tumor sections inFFPE were subjected to antigen retrieval as above,followed by exposure to fluorescein isothiocyanate(FITC)-conjugated anti-hypoxyprobe-1 mouse antibody(1:50) overnight at 4�C. This was detected with HRP-conjugated anti-FITC antibody (1:500, Jackson Immu-noresearch) and DAB resolution. Endogenous murine

Targeting ALDH1A1 Stem Cells in Ovarian Cancer

www.aacrjournals.org Mol Cancer Ther; 9(12) December 2010 3187




IgG was then blocked with anti-mouse IgG F(ab’)2fragments (Jackson ImmunoResearch), and ALDH1A1stained as above using AP-conjugated anti-mouse IgGand Ferangi Blue chromagen.

ALDEFLUOR assay and tumorigenicityin limiting dilutions

Active ALDH1A1 was identified with the ALDE-FLUOR assay according to manufacturer’s instructions(StemCell Technologies). The ALDH1A1-positive popu-lation was defined by cells with increased FITC signal,with gates determined by diethylaminobenzaldehyde(DEAB)-treated cells (DEAB being an inhibitor ofALDH1A1 activity). For tumorigenicity experiments,the ALDEFLUOR-positive population from A2780cp20cells were sorted with a FACS Aria II flow cytometer (BDBiosciences) and reanalyzed to confirm at least 95%positivity. Collected cells were washed and resuspendedin Ca2þ- and Mg2þ-free Hanks’ balanced salt solution(HBSS; Gibco) and injected intraperitoneally into NOD-SCIDmice in limiting dilutions . Mice were followed for 1year or until tumors formed, then sacrificed and tumorconfirmed histologically. For flow cytometric analysis ofthese tumors, xenografts were dissociated mechanicallywith a scalpel, passed through a 70-mm filter to collectsingle-cell suspensions, with the remaining clumped cellsincubated in 0.5 mg/mL of collagenase and 0.0369 mg/mL of hyaluronidase (Calbiochem) for 30 minutes at37�C. These chemically digested cells were again filteredthrough a 70 mm filter, added to the initial collectionand subjected to the ALDEFLUOR assay. ALDEFLUOR-positive cells or negative cells were then injected intoadditional mice (n ¼ 5) to examine maintenance oftumorigenicity.

Primary xenograft developmentWith institutional IRB and IACUC approval, excess of

freshly collected omental metastases from advancedstage ovarian cancer patients were acquired after tissuerequired for diagnosis and management had beensequestered. 3 to 4-mm3 sections were cut and implantedsubcutaneously on the dorsal aspect of NOD-SCID mice.Adjacent sections were submitted for histologic analysisto confirm tumor. Tumors were measured in 2 dimen-sions twice per week. After progressive growth wasnoted, mice with formed tumors were treated with vehi-cle or cisplatin (7.5 mg/kg weekly by intraperitonealadministration). Mice were treated for 8 weeks and thensacrificed, and tumors were harvested.

SiRNA downregulation in vitroTo examine downregulation of ALDH1A1with siRNA,

cells were exposed to 2.5 mg/mL of control siRNA (targetsequence 50-AATTCTCCGAACGTGTCACGT-30; Sigma),or 1 of 3 tested ALDH1A1-targeting constructs(SASI_Hs01_00244055, 00244056, or 00303091; Sigma),at a 1:3 siRNA (mg) to Lipofectamine 2000 (mL) ratio.

Lipofectamine 2000 and siRNA were incubated for20 minutes at room temperature, added to cells inserum-free RPMI to incubate for 6 hours, followed bythe addition of 15% FBS/RMPI thereafter. Transfectedcells were grown at 37�C for 48 to 72 hours and thenharvested for Western blot.

Assessment of cell viability with chemotherapyIC50 and cell-cycle analysis

To a 96-well plate, 2,000 cells per well were exposedto increasing concentrations of docetaxel or cisplatinin triplicates. Viability was assessed by 2-hour incuba-tion with 0.15% MTT (Sigma) and spectrophotometricanalysis at OD450 (optical density at 450 nm). Foreffects of siRNA on IC50, cells were incubated withsiRNA for 24 hours in 6-well plates and then replatedin 96-well plates, and chemotherapy was administeredafter 12 hours to allow attachment. IC50 was determinedby finding the dose at which the drug had 50% ofits effect and calculated by the following equation:IC50 ¼ [(OD450max � OD450min)/2) þ OD450min]. Testof synergy was according to the Loewe additivity model(27) and calculated by the following equation: combina-tion index (CI) ¼ [D1/Dx1] þ [D2/Dx2] (where a CI of1 suggests an additive effect, <1 suggests synergy, and>1 suggests antagonism). For cell-cycle analysis,cells were transfected with siRNA as described pre-viously for 72 hours, trypsinized, washed in PBS,and fixed in 75% ethanol overnight. Cells were thencentrifuged, washed twice in PBS, and reconstitutedin PBS with 50 mg/mL of propidium iodide. Propidiumiodide fluorescence was assessed by flow cytometry,and percentage of cells in each cycle was calculated bythe cell-cycle analysis module for FlowJo.

Orthotopic ovarian cancer model and in vivodelivery of siRNA

For orthotopic therapy experiments using ovarian can-cer cell lines, female athymic nude mice (NCr-nu) werepurchased from the National Cancer Institute and caredfor in accordance with guidelines of the American Asso-ciation for Accreditation of Laboratory Animal Care. Forall in vivo experiments, trypsinized cells were suspendedin HBSS and 106 cells injected intraperitoneally into 40mice per experiment. After 1 week, mice were rando-mized to: a) control siRNA/DOPC, b) control siRNA/DOPC plus chemotherapy, c) ALDH1A1-targetingsiRNA/DOPC, or d) chemotherapy plus ALDH1A1-tar-geting siRNA/DOPC. SiRNA/DOPC dose was 5 mgtwice per week in a volume of 100 mL intraperitoneally.Chemotherapy doses were docetaxel 35 mg intraperito-neally weekly for SKOV3TRip2, or cisplatin 160 mg intra-peritoneally weekly for A2780cp20. Mice were treated for4 weeks before sacrifice and tumor collection. SiRNAwasincorporated into 1,2-dioleoyl-sn-glycero-3-phosphati-dylcholine (DOPC) neutral nanoliposomes as previouslydescribed (28), lyophilized, and reconstituted in 0.9%saline for administration.

Landen et al.

Mol Cancer Ther; 9(12) December 2010 Molecular Cancer Therapeutics3188




Statistical analysisComparisons between treatment groups of tumor

weight was carried out with the 2-tailed Student’s t test,if tests of data normality were met. Those represented byalternate distribution were examined by Mann–WhitneyU statistic. Differences between groups were consideredstatistically significant at P < 0.05. The number ofmice pergroup (n¼ 10)was chosen as directed by a power analysisto detect a 50% decrease in tumor growth with b error of0.2. Progression-free and overall survival in patients with3 categories of ALDH1A1 staining were compared byplottingwith the Kaplan–Meier method and assessing forstatistical differences with the log-rank statistic, usingPASW 17.0 software.

Results

Expression profiling of chemoresistant ovariancancer cell linesTo discover genes mediating taxane resistance, expres-

sion profiling of parental SKOV3ip1 and taxane-resistantSKOV3TRip2 cells was done with microarray analysisusing the Illumina HumanRef-8 Expression BeadChip.The SKOV3TRip2 cell line was previously generatedthrough progressive exposure to paclitaxel (designatedSKOV3TR; 23) and then passaged intraperitoneally inmice for 2 generations to select populations withenhanced tumorigenicity. Similarly, SKOV3ip1 werederived from SKOV3 parental cells to select for cells withenhanced tumorigenicity. We found 34 genes to be upre-gulated more than 10-fold in SKOV3TRip2 (Fig. 1),among which was ALDH1A1, with a 92.7-fold increase(P ¼ 0.0025). Twenty genes were more than 10-foldincreased in SKOV3ip1. SKOV3TRip2 cells were con-firmed to have approximately 3,000-fold increased resis-tance to docetaxel, as measured by MTT IC50 (62.5 nmol/L vs. 0.02 nmol/L; Fig. 2A).

ALDH1A1 expression in ovarian cancer cell linesTo confirm an increase in ALDH1A1 expression/activ-

ity in SKOV3TRip2 and examine expression in otherovarian cancer cell lines, 4 pairs of parental and chemore-sistant cell lines were examined: SKOV3ip1/SKOV3-TRip2; HeyA8/HeyA8MDR (multidrug resistant);A2780ip2/A2780cp20 (10-fold increased cisplatin resis-tance); and IGROV-AF1/IGROV-cp20 (5-fold increasedcisplatin resistance). In addition, an immortalized, non-transformed cell line derived from normal ovarian sur-face epithelium, HIO-180, was examined. We found thatexpression of total ALDH1A1, as measured by Westernblot analysis, was in each case higher in the chemoresis-tant cell line, with the exception of HeyA8/HeyA8MDR,in which ALDH1A1 was low to absent in both (Fig. 2B).To examine whether ALDH1A1 was not only present butalso active, we subjected cells to flow cytometric analysisusing the ALDEFLUOR assay. This functional assaypredominantly identifies active ALDH1A1 by conversionof a chemical to a fluorochrome. The presence of a sub-

population of ALDH1A1-active cells could be readilyidentified in SKOV3TRip2 (58% of the total population)and A2780cp20 (2.2%) but not in their parental cell line(Fig. 2C). Furthermore, the strong shift in fluorescentsignal in some cells suggests that there was not simplya general increase in expression in all cells but ratherseparate populations of ALDH1A1-positive and -nega-tive cells. This was confirmed by immunohistochemistry,which showed distinct populations of ALDH1A1-posi-tive or -negative cells in A2780cp20 and SKOV3TRip2cells but not in the parental A2780ip2 and SKOV3ip1 cellsin culture (Fig. 2D). Finally, we observed that this hetero-geneous profile was maintained in tumors. After intra-peritoneal injection of SKOV3TRip2 cells into nude miceand collection of the resulting orthotopic tumor implants,IHC staining of for ALDH1A1 showed both positive andnegative ALDH1A1 subpopulations (Fig. 2E). To examinewhether this heterogeneity in expression was due todifferential expression in hypoxic regions, a tumor-bear-ing mouse was injected with hypoxyprobe reagent andsacrificed after 30minutes. The tumor was costainedwithALDH1A1 and antihypoxyprobe antibody. We foundthat the ALDH1A1-positive cells were not preferentiallylocalized to hypoxic regions in the tumor, with only 1.5%of ALDH1A1-positive cells concurrently positive forhypoxyprobe and only 3.3% of hypoxyprobe-positivecells also positive for ALDH1A1 (P < 0.01; Fig. 2F).

ALDH1A1 expression in human ovariancancer specimens

To determine the pattern of ALDH1A1 expression andpossible correlations with chemoresistance in patients,we next examined ALDH1A1 expression in 65 untreated,high-grade papillary serous stage III–IV ovarian cancerpatient specimens (patient characteristics in Table 1). Wefound a wide range of expression patterns (Fig. 3A).There was no ALDH1A1 in tumor cells in 27.1% ofsamples. ALDH1A1 expression was noted in 1% to20% of cells in 44% of tumors, representing the largestcohort of expression patterns. As in xenografts from celllines, expression was typically strong in some cellsand negative in others, signifying distinct heterogeneityin the tumor. There was no distinct histologic patternto the location of the positive cells (such as aroundvasculature or on the leading edge of the tumor), butpositive cells did tend to cluster together. The remainingtumors (28.9%) all had between 21% and 100% staining,with 10% of all patients having strong ALDH1A1expression in nearly 100% of their tumor cells. To confirmthat ALDH1A1 expression was not being mistakenlyidentified in tumor-infiltrating macrophages, severalsnap-frozen samples were dual stained for ALDH1A1and CD68. Although images are not as detailed as thosefrom paraffin-embedded samples, dual staining clearlyshows that the majority of macrophages (blue) areALDH1A1 negative and therefore the heterogeneousALDH1A1 positivity in tumors is not simply due todetection of macrophage infiltration (Fig. 3B).






Figure 1. Comparison of wholegenome expression profilingbetween SKOV3TRip2 andSKOV3ip1 cell lines. Total RNAfrom the SKOV3TRip2 andSKOV3ip1 cell lines weresubjected to whole genomeexpression profiling using theIllumina platform. The genes with agreater than 10-fold increase inSKOV3TRip2 are shown in red,whereas those with a greater than10-fold increase in SKOV3ip1 areshown in green. FC, fold change.

Landen et al.





Correlation of ALDH1A1 expressionwith clinical outcomesTodeterminewhetherALDH1A1expression correlated

with clinical outcomes, we compared progression-free

survival and overall survival from patient samplesdescribed earlier (and in Table 1) in cohorts with noALDH1A1 expression, 1% to 20% expression, and greaterthan 20% expression, as this grouping allowed similar

× ×

× ×

× ×

A B C

D

E

F

Figure 2. ALDH1A1expression in ovarian cancer cell lines. A, asmeasuredwith theMTTviability assay, theSKOV3TRip2ovarian cancer cell line has adocetaxelIC50 approximately 3,000-fold higher than that of its parental SKOV3ip1 cell line. B, expression of ALDH1A1 by Western blot in 4 pairs of chemosensitive andchemoresistantovariancancercell linesand thenontransformedHIO-180normalovariansurfaceepithelium line. Inall casesexceptHeyA8/HeyA8MDR, inwhichboth lines had minimal expression, the chemoresistant line had increased ALDH1A1 expression. C, as measured by the ALDEFLUOR assay, the A2780cp20(cisplatin resistant) and SKOV3TRip2 (taxane resistant) also contained a higher percentage of cells with functional ALDH1A1. D, this was confirmed by the IHCanalysis for ALDH1A1 on these cell lines in vitro in which individual cells appeared either negative or strongly positive, demonstrating heterogeneity of ALDH1A1expression in the cell line population. A low-power (4�) view gives an appreciation for the distinct colonies of ALDH1A1-positive cells, whereas examination athigh power (10�) shows the definitive ALDH1A1-positive or -negative nature of individual A2780cp20 and SKOV3TRip2 cells but an absence of ALDH1A1 inparentalA2780ip2andSKOV3ip1 lines.E, thisheterogeneity is alsopresent in tumorxenografts, asseenby the IHCanalysis forALDH1A1 inSKOV3TRip2 tumorsgrown inmice (intraperitoneal location isconfirmedby thepresenceofnormalpancreatic tissueon the right sideof theslide). F,ALDH1A1expression isnot limitedto hypoxic cells, as shown in xenografts collected frommice given the hypoxyprobe reagent and subjected to the co-IHC analysis for ALDH1A1 (in blue) and thehypoxyprobe by-product (in brown). Scale bars represent 50 mm in 10� views, 100 mm in 4� views (E, F).






numbers between groups. Patients with greater than 20%ALDH1A1-positive cells had a shorter median progres-sion-free survival (6.1 months) than those with 1% to 20%ALDH1A1-positive cells (8.2 months) or those with noALDH1A1-positive cells (13.8 months), which was statis-tically significant according to the log-rank test (P¼ 0.035;Fig. 3C). Overall survival, which reflects resistance tomultiple chemotherapeutic agents used in the recurrentsetting, showed a trend toward a poor outcome withincreasing ALDH1A1 expression (median overall survi-val 1.09 vs. 1.84 vs. 2.32 years), but the trend was notstatistically significant (P ¼ 0.33; Supplementary Fig. 1).

Preferential survival of ALDH1A1-positivecells with cisplatin treatment

To determine whether the ALDH1A1-positive cellshavepreferential survival in the tumormicroenvironmentwith platinum treatment, we established mouse xeno-grafts from primary patient samples by subcutaneouslyimplanting a freshly collected tumor specimen into NOD-SCIDmice. A subcutaneous rather than orthotopic modelwas used so that tumor growth and response could beaccurately measured. Once tumors were established and

growing, and achieved a size of approximately 1 cm3,intraperitoneal administration of 7.5 mg/kg of cisplatinweekly was initiated whereas only vehicle was adminis-tered to controls (Fig. 3D). When tumors grew to a size of2 cm3 in controls, having remained stable with cisplatintreatment, they were harvested and sections stained forALDH1A1 expression. Baseline expression of ALDH1A1in the implanted tumor was seen in approximately 1% ofcancer cells and similar levels were found in growingxenografts in untreated mice (Fig. 3E). A significantincrease in the percentage of ALDH1A1-positive cellswas, however, noted in cisplatin-treated xenografts to38% (P < 0.001; Fig. 3E). Consistent with this, the ALDE-FLUOR assay on the dissociated tumor showed that 0.6%of cells from untreated tumors were ALDEFLUOR posi-tive whereas 17.6% of cells from cisplatin-treated tumorswere ALDEFLUOR positive. Because the treated xeno-graft in this case did not regress, but rather remainedstable in size, cisplatin exposure may have inducedALDH1A1 expression in surviving cells in addition topreferential killing of ALDH1A1-negative cells.

Tumor-initiating capacity of ALDH1A1-positiveovarian cancer cells

In breast and other cancers, the ALDH1A1-active can-cer cells have been shown to represent a tumor-initiatingpopulation (10–19). To determine whether this were thecase in ovarian cancer, we sorted ALDH1A1-positive and-negative populations from the A2780cp20 cell line usingthe ALDEFLUOR assay and injected cells intraperitone-ally into NOD-SCID mice in limiting dilutions to deter-mine tumor-initiating potential. As summarized inTable 2, ALDEFLUOR-positive cells exhibited increasedtumorigenic potential, with 100% tumor initiation afterthe injection of 100,000, 25,000, or 5,000 cells, and 1 tumorwas established after the injection of 1,000 cells. ALDE-FLUOR-negative cells could form tumors, although at alower rate: 2 of 5mice formed tumors after the injection of25,000 or 100,000 cells and no tumors formed after theinjection of 5,000 or 1,000 cells. Mice were followed for1 year after injection and thorough necropsies wereperformed in remaining mice to confirm that tumorsfailed to develop. The TD50, or dose of cells required topermit tumor formation in 50% of animals, was 50-foldlower with ALDEFLUOR-positive cells. Perhaps, morestriking was the makeup of these tumors. One require-ment of a tumor-initiating population is that it has thecapacity to give rise to heterogeneous tumors, composedof both stem cell and non–stem cell populations, therefore

Table 1. Characteristics of patients tested forALHD1A1 expression (n ¼ 65)

Characteristic Percentage oraverage (range)

Age at diagnosis 62.2 (34–89)Caucasian race 71%Pretreated with chemotherapy 0%Stage

III 74%IV 26%

Ca125 3,071 (161–9,600)Ascites 87%Optimal debulking 74%Papillary serous histology 100%Platinum/taxane primary therapy 96%Progression-free survival, mo 14.2 (1.7–108)Overall survival, y 2.5 (0.2–11.8)ALDH1A1 staining

Absent 27.1%1%–20% of cells 44.0%21%–100% of cells 28.9%

Abbreviation: Ca125, cancer antigen 125

Table 2. Tumorigenicity of ALDEFLUOR-positive and negative cells

A2780cp20 cells injectedintraperitoneally

1,000,000 250,000 100,000 25,000 5,000 1,000 Serialtransplantation rate

ALDEFLUOR negative 5/5 4/5 2/5 2/5 0/5 0/5 0/5ALDEFLUOR positive 5/5 5/5 5/5 1/5 5/5

Landen et al.





demonstrating multipotent differentiation potential. Thiswas noted in tumors that formed after the injection ofALDEFLUOR-positive cells. In all 16 of these tumors, astrongly positive ALDH1A1 population was noted in theminority of the sample, on average 4.7% of the tumor(range 2.4%–6.1%; Fig. 4A). However, no ALDEFLUOR-positive cells were found in the tumors that formed afterthe injection of ALDH1A1-negative cells (Fig. 4B). Thiswas confirmed by the IHC analysis (Fig. 4C and D). Thisargues against the idea that tumors are formed because ofcontamination with ALDEFLUOR-positive cells or thatALDH1A1 expression is simply induced by the tumormicroenvironment regardless of the capacity of the cells.

This difference in the capacity to generate ALDEFLUOR-positive cells was also noted in vitro. SKOV3TRip2 cellssorted into ALDEFLUOR-positive and -negative popula-tions were cultured separately, and the ALDEFLUORassay was done on the different populations at 24, 48,and 72 hours (Fig. 4E and F). Of the ALDEFLUOR-positive cells, the population gradually reverted to75.3%, 54.2%, and 51.4% ALDEFLUOR-positive cells,respectively, for each time point. However, the ALDE-FLUOR-negative cells could not produce any ALDE-FLUOR-positive cells.

To confirm that the ALDEFLUOR-positive cells fromtumors maintained tumorigenicity, these populations

A

D

E

B C

Figure 3. ALDH1A1 expression in ovarian cancer patients. ALDH1A1 was assessed by the IHC analysis in 65 high-grade stage III–IV papillary serous ovariancancer patients. A, several expression patterns were seen, including absent, spotty (e.g., Low ALDH), and diffuse (High ALDH) staining. Consistent withstaining in cell lines, both strongly positive and negative populations were noted. B, to confirm the spotty ALDH1A1 pattern was not identifying infiltratingmacrophages, the co-IHC analysis on frozen tissue for ALDH1A1 (brown) and CD68 (a pan-macrophagemarker, blue) was done. C, patients were stratified intoless than 1%, 1%–20%, and greater than 20% ALDH1A1 expression, and progression-free and overall survival was plotted by the Kaplan–Meier method andtested for statistical significance by the log-rank test. There was a significantly shorter progression-free survival in patients with increasing ALDH1A1expression. D, mice with established primary subcutaneous xenografts were treated with vehicle or cisplatin for 5 weeks. E, tumors from these mice wereharvested and subjected to IHC analysis for ALDH1A1. Tumors treated with cisplatin showed a significant increase in the number of ALDH1A1-positive tumorscells. Magnification at low and high powers is shown. Scale bars represent 50 mm in panels A, B, and High-power images of E, and 100 mm in Low-powerimages of E.






were sorted and reinjected intraperitoneally into miceand continued to form tumors at 100% rate with 25,000cells injected. However, ALDEFLUOR-negative cellsfrom the tumors forming after ALDEFLUOR-negativecells were injected did not form tumors. Taken together,these studies show that ALDEFLUOR-positive cells haveincreased but not absolute tumorigenicity, but they dohave a differentiation capacity and maintenance of thetumorigenic phenotype that is absent in ALDEFLUOR-negative cells.

In an effort to determine whether ALDEFLUOR-posi-tive cells, freshly collected from ovarian cancer patients,have similar tumorigenicity, we have sorted ALDE-

FLUOR-positive and -negative cells from 5 separate ovar-ian cancer patients, dissociating tumors metastatic to theomentum at the time of primary debulking surgery. Inthis cohort, 1.5% to 17.8% of cells were ALDEFLUORpositive. A total of 25,000 ALDEFLUOR-positive cells,100,000 ALDEFLUOR-negative cells, or 100,000 unsortedcells were injected intraperitoneally into 5mice per groupper patient. Unfortunately, no tumors formed in anymice, highlighting the difficulty of tumorigenicity studiesin primary ovarian cancer samples dissociated to singlecell suspensions.

To preliminarily determine whether there is an overlapbetween the ALDEFLUOR-positive population and other

Tumors from ALDH-positive cells

Control (+DEAB)256

R3 R3 R3 R3

192

128

64

010

010

110

210

310

4

100

101

102

103

104

100

101

102

103

104

100

101

102

103

104

100

101

102

103

104

100

101

102

103

104

100

101

102

103

104

100

101

102

103

104

100

101

102

103

104

100

101

102

103

104

256

SS

Are

a

SS

Are

a

Sid

e S

catte

rS

ide

Sca

tter

Sid

e S

catte

rS

ide

Sca

tter

192

128

64

0

256

192

128

64

0

256

192

128

64

0

FITC (ALDEFLUOR)

FITC (ALDEFLUOR) FITC (ALDEFLUOR)

24 hrs 48 hrs 72 hrs 24 hrs 48 hrs 72 hrs

FITC (ALDEFLUOR)FITC (ALDEFLUOR) FITC (ALDEFLUOR)

Control (+DEAB)ALDEFLUOR ALDEFLUOR

Tumor from ALDH-positive cells

Culture of ALDH-positive cells

Tumors from ALDH-Negative cells

Tumor from ALDH-Negative cells

Culture of ALDH-negative cells

1K

800

ALDEFLUOR75.3%

ALDEFLUOR54.2%

ALDEFLUOR51.4%

ALDEFLUOR0.738%

ALDEFLUOR0.050%

ALDEFLUOR0.946%

0.74%51.4%54.2%75.3% 0.05% 0.95%

600

400

200

0

1K

800

600

400

200

0

1K

800

600

400

200

0

1K

800

600

400

200

0

1K

800

600

400

200

0

1K

800

600

400

200

0

A B

C D

E F

Figure 4. ALDH1A1 populations in A2780cp20 xenografts. Intraperitoneal tumors that developed after the injection of ALDEFLUOR-positive or -negativeA2780cp20 cells were assessed for ALDH1A1 composition. A, tumors that formed after the injection of purely ALDEFLUOR-positive cells showed bothALDEFLUOR-positive and -negative populations and recapitulated the tumor-initiating cells phenotype of having a small (2.4%–6.1%) percentage ofALDEFLUOR-positive cells. B, interestingly, tumors that formed after the injection of purely ALDEFLUOR-negative cells contained only ALDEFLUOR-negativecells, showing an absence of capacity for differentiation, at least in terms of ALDEFLUOR positivity. C and D, this expression discrepancy was also noted onthe immunohistochemical analysis for ALDH1A1 from these samples. Scale bars represent 100 mm. Similarly, in vitro, SKOV3TRip2 ALDEFLUOR-positive cellsgive rise to both ALDEFLUOR-positive and -negative cells, (E) reestablishing baseline levels at 48 hours, whereas ALDEFLUOR-negative cells cannot give riseto ALDEFLUOR-positive cells (F).

Landen et al.





markers of putative stem cells in ovarian cancer, these5 samples were also profiled for CD44, c-kit, and CD133.We were not able to identify a convincing positive c-kitpopulation from any sample. CD133-positive cells madeup an average of 3.1% of total tumor cells (range, 0.6%–5.7%) and were greater than 80% of ALDEFLUOR posi-tive in all 5 samples (mean, 86.7%; range, 81.5%–100%).CD44 was more commonly expressed, representing anaverage of 45.7% of tumors (but with a very broad rangeof 2.4%–98.2%). Of the CD44-positive cells, 75.4% werealso ALDEFLUOR positive (range, 46.6%–88.8%). Simi-larly, the SKOV3TRip2 line has 82% CD44-positive cells,and of these, 74% were ALDEFLUOR positive. Althougha great number of samples will need to be examined tofully delineate whether multiple marker–positive cellscan more accurately define the most pure tumorigeniccell, there is certainly overlap inmarker expression. Thereare both double-positive CD44/ALDEFLUOR andCD133/ALDEFLUOR-positive populations that mayprove more discerning as cancer stem cell populations,and ongoing studies could assess this distinction. Inter-estingly, the A2780cp20 cell line is completely negativefor CD44 and the HeyA8 cell line is negative forALDH1A1/ALDEFLUOR, despite the fact that both arehighly tumorigenic. This highlights the fact that thesecannot be the sole mediators of tumorigenicity in mice.

Downregulation of ALDH1A1 sensitizes ovariancancer cells to chemotherapyGiven the association of ALDH1A1 expression with

chemoresistant cell lines and a decreased progression-free survival in ovarian cancer patients, we askedwhether downregulation of ALDH1A1 could sensitizeresistant cells to chemotherapy. Two different siRNAconstructs were identified that reduced ALDH1A1expression by greater than 80% (Fig. 5A). Reduction inthe ALDEFLUOR population was confirmed (Fig. 5B).SKOV3TRip2 or A2780cp20 cells were exposed toALDH1A1-targeting siRNA (ALDH1A1 siRNA) or con-trol siRNA for 24 hours before replating and addingincreasing concentrations of docetaxel or cisplatin,respectively. Cell viability 4 days after the addition ofchemotherapy was assessed with the MTT assay. InSKOV3TRip2 cells, siRNA-ALDH1A1 alone reduced via-bility by 49% (Fig. 5C; P < 0.001). Downregulation ofALDH1A1 also reduced the docetaxel IC50 from 178 to 82nmol/L. In A2780cp20, the effects of ALDH1A1 down-regulation alone were modest (Fig. 5D; reduced viabilityby 15.9%, P ¼ 0.040) but sensitization to cisplatin wasconsiderable, with a decrease in the IC50 from 5.1 to 2.0mmol/L. Tests for synergy suggest moderate synergy ineach cell line (CI ¼ 0.82 for SKOV3TRip2 and 0.75 forA2780cp20). The contrasting effects of ALDH1A1-siRNAalone are consistent with the number of ALDH1A1-activecells in these cell lines, with SKOV3TRip2 cell lineshaving 50% to 60% of ALDEFLUOR-positive cells andA2780cp20 having just 2% of 3%. To determine howALDH1A1 downregulation alone may affect cell growth,

cell-cycle analysis was done in a separate experiment.We found that ALDH1A1 downregulation induced anaccumulation of SKOV3TRip2 cells in S and G2 phases(P < 0.001; compared with control siRNA) but had onlyminimal effects on the cell cycle of A2780cp20 cells(Fig. 5E).

There are no known inhibitors of ALDH1A1 for in vivostudies. Therefore, we used a method for delivery ofsiRNA in vivo, using DOPC nanoparticles. We and others(28–32) have previously shown delivery of siRNA incor-porated into DOPC nanoliposomes to the tumor parench-yma with subsequent target downregulation. In thisstudy, nude mice were injected intraperitoneally witheither SKOV3TRip2 or A2780cp20 cells and randomizedto 4 treatment groups to begin 1 week after cell injection:a) control siRNA in DOPC, delivered intraperitoneallytwice per week; b) docetaxel 35 mg, delivered intra-peritoneally weekly (for SKOV3TRip2model) or cisplatin160 mg, delivered intraperitoneally weekly (forA2780cp20 model); c) ALDH1A1-siRNA in DOPC, intra-peritoneally twice per week; or d) ALDH1A1-siRNA inDOPC plus docetaxel (for SKOV3TRip2) or cisplatin (forA2780cp20). After 4 weeks of treatment, mice were sacri-ficed and total tumor weight recorded. The IHC analysisconfirmed reduced ALDH1A1 expression withALDH1A1-siRNA/DOPC treatment compared with con-trols but not with chemotherapy alone (SupplementaryFig. 2; too little tissue was available to examine with theALDEFLUOR assay). In SKOV3TRip2 xenografts(Fig. 5F), there was a nonsignificant reduction of 37.0%in tumor growth with docetaxel treatment (P ¼ 0.17) andof 25.0% with ALDH1A1 siRNA treatment (P ¼ 0.38)compared with control siRNA/DOPC. The observationthat ALDH1A1 downregulation alone significantlydecreased SKOV3TRip2 growth in vitro but was lesspronounced in vivo suggests that tumor microenviron-ment factors such as supporting stromal cells may be ableto protect cells from ALDH1A1 depletion. However, thecombination of ALDH1A1 siRNA and docetaxel resultedin significantly reduced growth by 93.6% compared withcontrol siRNA (P < 0.001), by 89.8% compared withdocetaxel plus control siRNA (P ¼ 0.003), and by91.4% compared with ALDH1A1 siRNA (P ¼ 0.002). InA2780cp20 (Fig. 5G), there was a similar nonsignificantreduction of 43.9% in tumor weight with cisplatin alone(P ¼ 0.32) and of 57.0%with ALDH1A1 siRNA treatment(P ¼ 0.19). These effects may be even less significant thanthe mean tumor weights suggest, given the presence of 2especially large tumors in the control siRNA group.However, again combined therapy showed a sensitiza-tion to chemotherapy with ALDH1A1 siRNA, with com-bination therapy reducing growth by 85.0% comparedwith control siRNA (P ¼ 0.048), by 73.4% compared withcisplatin plus control siRNA (P ¼ 0.013), and by 65.3%compared with ALDH1A1 siRNA alone (P ¼ 0.039).Given the minimal effects of each single agent and theconsistent finding of significant improvement with com-bined therapy, these data suggest a synergy between






ALDH1A1 downregulation and both taxane and plati-num chemotherapeutic agents, though formal dose-find-ing experiments would be required to definitively provesynergy.

Discussion

We have found that ALDH1A1 expression and activityare increased in chemoresistant ovarian cancer cell lines

A

C

D

E

G

F

B

Figure 5. Efficacy of ALDH1A1 downregulation with siRNA in vitro and in vivo. Identification of siRNA constructs that decrease ALDH1A1 expression wasconfirmed by Western blotting (A) and by flow cytometry (B) using the ALDEFLUOR assay in the SKOV3TRip2 cell line. C, downregulation of ALDH1A1 withsiRNA 48 hours prior to the treatment of SKOV3TRip2 cells with increasing doses of docetaxel showed a sensitization effect, decreasing IC50 from 178 to82 nmol/L. siRNA alone also showed an effect, with decreased viability by 49%. D, in the A2780cp20 cell line, downregulation of ALDH1A1 alone had minimaleffect but sensitized cells to cisplatin, decreasing IC50 from 5.1 to 2.0 mmol/L. E, cell-cycle analysis shows that ALDH1A1 downregulation inducesaccumulation of cells in S andG2 phases in SKOV3TRip2, with little effect on A2780cp20. F, in vivo, mice injected intraperitoneally with SKOV3TRip2 cells weretreated with ALDH1A1-siRNA incorporated in DOPC nanoparticles, docetaxel/control siRNA in DOPC, or the combination and compared with mice treatedwith control siRNA in DOPC. Mice treated with either of the single agents had minimal effect, but the combination showed a significant reduction comparedwith treatment with control siRNA (94% reduction in tumor growth; P < 0.001) or either of the single agents (90%–91% reduction; P < 0.005). G, similarly, miceinjected with A2780cp20 cells showed a minimal, nonsignificant reduction in growth with cisplatin or ALDH1A1-siRNA in DOPC, but combination therapy wasstatistically superior to either of the single agents (65%–73% reduction; P < 0.04) or control siRNA (85% reduction; P ¼ 0.048). Mean tumor weight andindividual tumor sizes are presented.

Landen et al.





and in in situ primary ovarian cancer xenografts treatedwith cisplatin. Expression of ALDH1A1 is frequent inovarian tumors, and patients with lowALDH1A1 expres-sion levels have a more favorable outcome than thosewith more ALDH1A1-positive cells. ALDEFLUOR-positive cells have increased (but not absolute) tumor-igenicity compared with ALDEFLUOR-negative cellsand have a differentiating capacity that is not presentin the ALDEFLUOR-negative population. Most impor-tant, downregulation of ALDH1A1 expression sensitizednormally chemoresistant tumors to both docetaxel andcisplatin both in vitro and in an orthotopic mouse modelof ovarian cancer.The search for tumor-initiating cells in ovarian cancer

has resulted in observations that the CD44þ/c-kitþ popu-lation has an approximately 5,000-fold increase in tumor-igenicity, with tumors forming after the injection of as fewas 100 cells from primary tumor, xenograft, or spheroidheterogeneous populations (5), and that the CD133þ

population has approximately 20-fold increased tumor-igenicity, with tumor formation with as few as 100 to 500cells from murine xenografts, and tumor formation 4times faster with CD133þ cells (7). Furthermore, theincreased tumorigenicity of CD133þ cells can be inhibitedby interfering with binding between CD44 and its ligandhyaluronic acid (6). Other investigators have found equalrates of tumor formationamongCD133þandCD133� cellsfromtheA2780 cell line, but a fastergrowth rate inCD133þ

cells (8). The side population (SP) cells from theMOVCARcell line also formed tumors more frequently andappeared 3 to 4 weeks sooner than tumors derived fromnon-SP cells (9). In all of these studies, as in ours, thetumors resulting from the putative tumor-initiating cellpopulation contained both tumor-initiating cell and non-tumor-initiating cell populations, demonstrating multi-potentiality. Interestingly, we have seen that cells com-prising tumors formed from ALDH1A1-negative cellslack the capacity to generate ALDH1A1-positive cellsand do not continue to propagate tumors over multiplegenerations, suggesting that their multipotentiality islimited. This lack of differentiating capacity has also beennoted in ALDEFLUOR-negative cells from breast cancercell lines (33).The most appropriate source of tumor cells for tumor-

igenicity experiments is of some debate. Although it isdesirable to use samples freshly collected from primarytumors, sorting these samples and establishing primaryxenografts have proven problematic. Ovarian cancerxenografts and cells lines have traditionally been chal-lenging to establish from primary samples. All pre-viously reported studies of ovarian tumor-initiatingcells have used selected cells of some sort, either fromxenografts of varying generations or from cells grownin differentiation-inhibiting media (to form tumorspheres), to serve as a compromise between freshlycollected specimens and cell lines. However, those cellsthat form tumors in mice even in the first generationalmost certainly represent some select portion of the

original tumor. That these xenografts still contain only asmall percentage of tumor-initiating cells speaks eitherto the appropriateness of this approach or to the testa-ment that the tumor-forming cells are multipotent, giverise to tumor-initiating cell-negative populations, andremain relatively rare. Use of cell lines is often discour-aged because of their homogenous nature. But clearly,even within cell lines, there is heterogeneity inALDH1A1 expression, as shown by the detection ofdistinct populations by flow cytometric and IHC ana-lyses (Fig. 2). Distinct ALDEFLUOR-positive and -nega-tive populations have also been found in several breastcancer cell lines, with ALDEFLUOR-positive cells hav-ing increased tumorigenicity and differing molecularsignatures (33). Therefore, our finding that the ALDE-FLUOR-positive population in cell lines has increasedtumorigenicity may reflect the more aggressive pheno-type of ALDH1A1-active cells but does not representproof that this population is important to in situ ovariancancers. Evidence that patients with increasingALDH1A1 expression have poor outcomes suggests thisassociation, but additional tumorigenicity experimentsfrom freshly collected tumors would more appropri-ately define the ALDEFLUOR population as clinicallysignificant tumor-initiating cells.

The importance of tumorigenicity in defining cancerstem cells has also been debated. Although tumor for-mation with 100 to 500 ALDEFLUOR-positive cells and alack of tumor formation with the injection of 105 ALDE-FLUOR-negative cells definitely reflect an aggressivephenotype, the biologic processes required for xenograftformation-–survival under stressful experimental condi-tions, adhesion, time to proliferation, and variations inhost immunocompetence-–may not reflect the true popu-lation that cancer stem cell research seeks to identify. Ourultimate goal should be to identify the subpopulations inparent tumors that survive chemotherapy and thereforeare more likely to cause recurrence. Stem cells that sur-vive chemotherapy should exhibit chemoresistance to beclinically relevant. In breast cancer, for example, theCD44þ/CD24� population is highly tumorigenic. How-ever, Tanei et al., who studied tissue obtained before andafter neoadjuvant chemotherapy, found that despite apositive response to treatment, the proportion of CD44þ/CD24�-negative cells was unchanged. In these samples,however, the ALDH1A1-positive population was signif-icantly increased (34).

ALDH1A1 has previously been proposed to play a rolein chemoresistance, having been noted to be higher inproteomic profiling of IGROV platinum-resistant ovariancancer cells (35), in genomic profiling of multidrug-resis-tant gastric carcinoma (36), and in cells resistant to cyclo-phosphamide (37, 38), oxazaphosphorines (39), and nowdocetaxel and cisplatin. ALDH1A1 oxidizes many intra-cellular aldehydes into carboxylic acids (40), detoxifyingmany of the free oxygen radicals generated by chemother-apeutic agents. It stands to reason that a stem cell popula-tion should be resistant to multiple chemotherapeutic






agents rather than being specific to one class. This alsofollows clinically, in that most ovarian cancer patientswho develop resistance to platinum agents have resis-tance tomultiple agents (2). ALDH1A1 has been shown tobe associated with BRCA1 in breast cancer, in that knock-down of BRCA1 increases the ALDEFLUOR populationand ALDEFLUOR-positive cells preferentially containBRCA1 loss of heterozygosity (41). These findings couldalso be important to BRCA-mediated ovarian cancer.Despite this body of evidence for the importance ofALDH1A1, it is not fully understood whether any ofthe additional ALDH1 isoforms are important to stemcell biology. In our study, ALDH1A1 can be specificallyidentifiedwith isotype-specific antibodies (as used for theIHC analysis and Western blotting). However, the moreimportant and consistently used identifier of a stem cellpopulation is the ALDEFLUOR assay, which, althoughprimarily dependent on ALDH1A1, may also identifyALDH1A2andALDH1A3 isotypes [(42) andunpublisheddata by Stem Cell Technologies]. As a therapeutic agent,we have seen positive effects by targetingALDH1A1withsiRNA, but to maximize the efficacy of therapeutics, thecontribution of these additional isotypes will need to bedefined with additional studies.

Although our finding of a poor outcome in patientswith high ALDH1A1 expression agrees with similarinvestigations in breast cancer (12, 13) and ovarian cancer(20), one interesting report found that a high ALDH1A1expression level actually confers a positive prognosis inovarian cancer (43). This cohort also contained patientswith absent, scattered, and diffuse staining. However,this cohort included patients with stage I and II diseaseand low-grade tumors, and ALDH1A1 expression washigher in these patients [confirming findings from aprevious report (44)]. Furthermore, with multivariateanalysis, only stage correlated with survival; ALDH1A1expression no longer predicted outcomes. In ovariancancer, there is a well-recognized dichotomy in carcino-genesis and pathobiology (45), whereby low-grade

tumors (which are more often diagnosed at stage I orII) are paradoxically more chemoresistant but have pro-longed survival due to slow growth. Given these collec-tive data, and the several mechanisms by whichALDH1A1 has been shown to contribute to chemoresis-tance, it may be that ALDH1A1 is more frequentlyexpressed in low-grade tumors but participates in che-moresistance to both high-grade and low-grade subtypes.

We have shown that the ALDH1A1-positive popula-tion has properties of cancer stem cells, is associated withtaxane and platinum resistance, and can be resensitizedto chemotherapy with downregulation of ALDH1A1in vitro and in vivo. Therefore, ALDH1A1 is not just amarker of an aggressive population but also amediator ofthe phenotype and a viable target for therapy. As bettermodels are developed to more purely define the truechemoresistant population in de novo patient tumors, theALDH1A1 population, either alone or in combinationwith other markers and mediators of resistance, mayrepresent a population that must be targeted to achieveincreased response rates and survival in ovarian cancerpatients.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Grant Support

Funding support provided by the Reproductive Scientist DevelopmentProgram through the Ovarian Cancer Research Fund and the NationalInstitutes of Health (NIH; K12 HD00849), the Department of DefenseOvarian Cancer Research Academy, a Career Development Award fromthe UT MD Anderson Cancer Center Specialized Program of ResearchExcellence in ovarian cancer (CA083639), and the Gynecologic CancerFoundation to C.N.L.; NIH (CA109298 and CA110793, P50 CA083639;CA128797, RC2GM092599), Program Project Development Grant from theOvarian Cancer Research Fund, the Marcus Foundation, and the BettyAnn Ashe Murray Distinguished Professorship to A.K.S.

Received 06/16/2010; revised 09/03/2010; accepted 09/17/2010;published OnlineFirst 10/01/2010.

References1. Jemal A, Siegel R, Ward E, Hao Y, Xu J, Thun MJ. Cancer statistics,

2009. CA Cancer J Clin 2009;59:225–49.2. Hoskins W, Perez C, Young R, Barakat R, Markman M, Randall M.

Principles and Practice of Gynecologic Oncology. 4th ed.Philadelphia:Lippincott Williams & Wilkins; 2005.

3. Rosen JM, Jordan CT. The increasing complexity of the cancer stemcell paradigm. Science 2009;324:1670–3.

4. Dalerba P, Cho RW, Clarke MF. Cancer stem cells: models andconcepts. Annu Rev Med 2007;58:267–84.

5. Zhang S, Balch C, Chan MW, Lai HC, Matei D, Schilder JM, et al.Identification and characterization of ovarian cancer-initiating cellsfrom primary human tumors. Cancer Res 2008;68:4311–20.

6. SlomianyMG,Dai L, Tolliver LB,GrassGD, ZengY, TooleBP. Inhibitionof functional hyaluronan-CD44 interactions in CD133-positive primaryhuman ovarian carcinoma cells by small hyaluronan oligosaccharides.Clin Cancer Res 2009;15:7593–601.

7. Curley MD, Therrien VA, Cummings CL, Sergent PA, Koulouris CR,Friel AM, et al. CD133 expression defines a tumor initiating cell

population in primary human ovarian cancer. Stem Cells 2009;27(12):2875–83.

8. Baba T, Convery PA, Matsumura N, Whitaker RS, Kondoh E, Perry T,et al. Epigenetic regulation of CD133 and tumorigenicity of CD133þ

ovarian cancer cells. Oncogene 2009;28(2):209–18.9. Szotek PP, Pieretti-Vanmarcke R, Masiakos PT, Dinulescu DM,

Connolly D, Foster R, et al. Ovarian cancer side population definescells with stem cell-like characteristics and Mullerian Inhibiting Sub-stance responsiveness. Proc Natl Acad Sci U S A 2006;103:11154–9.

10. Moreb JS. Aldehyde dehydrogenase as a marker for stem cells. CurrStem Cell Res Ther 2008;3:237–46.

11. Balicki D. Moving forward in human mammary stem cell biology andbreast cancer prognostication using ALDH1. Cell Stem Cell2007;1:485–7.

12. Ginestier C, Hur MH, Charafe-Jauffret E, Monville F, Dutcher J, BrownM, et al. ALDH1 is a marker of normal and malignant human mammarystem cells and a predictor of poor clinical outcome. Cell Stem Cell2007;1:555–67.

Landen et al.





13. Charafe-Jauffret E, Ginestier C, Iovino F, Tarpin C, Diebel M, Esterni B,et al. Aldehyde dehydrogenase 1-positive cancer stem cells mediatemetastasis and poor clinical outcome in inflammatory breast cancer.Clin Cancer Res 2010;16:45–55.

14. Croker AK, Goodale D, Chu J, Postenka C, Hedley BD, Hess DA,et al. High aldehyde dehydrogenase and expression of cancerstem cell markers selects for breast cancer cells with enhancedmalignant and metastatic ability. J Cell Mol Med 2009;13(8B):2236–52.

15. Huang EH, Hynes MJ, Zhang T, Ginestier C, Dontu G, Appelman H,et al. Aldehyde dehydrogenase 1 is a marker for normal and malignanthuman colonic stem cells (SC) and tracks SC overpopulation duringcolon tumorigenesis. Cancer Res 2009;69:3382–9.

16. Carpentino JE, HynesMJ, Appelman HD, Zheng T, Steindler DA, ScottEW, et al. Aldehyde dehydrogenase-expressing colon stem cellscontribute to tumorigenesis in the transition from colitis to cancer.Cancer Res 2009;69:8208–15.

17. Dembinski JL, Krauss S. Characterization and functional analysis of aslow cycling stem cell-like subpopulation in pancreas adenocarci-noma. Clin Exp Metastasis 2009;26:611–23.

18. Ucar D, Cogle CR, Zucali JR, Ostmark B, Scott EW, Zori R, et al.Aldehyde dehydrogenase activity as a functional marker for lungcancer. Chem Biol Interact 2009;178:48–55.

19. Ma S, Chan KW, Hu L, Lee TK, Wo JY, Ng IO, et al. Identification andcharacterization of tumorigenic liver cancer stem/progenitor cells.Gastroenterology 2007;132:2542–56.

20. Deng S, Yang X, Lassus H, Liang S, Kaur S, Ye Q, et al. Distinctexpression levels and patterns of stem cell marker, aldehyde dehy-drogenase isoform 1 (ALDH1), in human epithelial cancers. PLoS One2010;5:e10277.

21. Buick RN, Pullano R, Trent JM. Comparative properties of fivehuman ovarian adenocarcinoma cell lines. Cancer Res1985;45:3668–76.

22. Yoneda J, Kuniyasu H, Crispens MA, Price JE, Bucana CD, Fidler IJ.Expression of angiogenesis-related genes and progression ofhuman ovarian carcinomas in nude mice. J Natl Cancer Inst1998;90:447–54.

23. Duan Z, Feller AJ, Toh HC, Makastorsis T, Seiden MV. TRAG-3, anovel gene, isolated from a taxol-resistant ovarian carcinoma cell line.Gene 1999;229:75–81.

24. Illumina I. BeadStudio Normalization Algorithms for Gene ExpressionData. Technical note: Illumina� RNA analysis 2007. Available from:http://www.illumina.com/Documents/products/technotes/technote_beadstudio_normalization.pdf (accessed July 1, 2010).

25. Benjamini Y, Drai D, Elmer G, Kafkafi N, Golani I. Controlling the falsediscovery rate in behavior genetics research. Behav Brain Res2001;125:279–84.

26. Landen CN Jr, Lu C, Han LY, Coffman KT, Bruckheimer E, Halder J,et al. Efficacy and antivascular effects of EphA2 reduction with anagonistic antibody in ovarian cancer. J Natl Cancer Inst2006;98:1558–70.

27. Straetemans R, O’Brien T, Wouters L, Van Dun J, JanicotM, Bijnens L,et al. Design and analysis of drug combination experiments. Biom J2005;47:299–308.

28. Landen CN Jr, Chavez-Reyes A, Bucana C, Schmandt R, Deavers MT,Lopez-Berestein G, et al. Therapeutic EphA2 gene targeting in vivousing neutral liposomal small interfering RNA delivery. Cancer Res2005;65:6910–8.

29. Gray MJ, Dallas NA, Van Buren G, Xia L, Yang AD, Somcio RJ, et al.Therapeutic targeting of Id2 reduces growth of human colorectalcarcinoma in the murine liver. Oncogene 2008;27:7192–200.

30. Halder J, Kamat AA, Landen CN Jr, Han LY, Lutgendorf SK, Lin YG,et al. Focal adhesion kinase targeting using in vivo short interferingRNA delivery in neutral liposomes for ovarian carcinoma therapy. ClinCancer Res 2006;12:4916–24.

31. Kamat AA, Feng S, Agoulnik IU, Kheradmand F, Bogatcheva NV,Coffey D, et al. The role of relaxin in endometrial cancer. Cancer BiolTher 2006;5:71–7.

32. Villares GJ, Zigler M, Wang H, Melnikova VO, Wu H, Friedman R, et al.Targeting melanoma growth and metastasis with systemic delivery ofliposome-incorporated protease-activated receptor-1 small interfer-ing RNA. Cancer Res 2008;68:9078–86.

33. Charafe-Jauffret E, Ginestier C, Iovino F, Wicinski J, Cervera N, FinettiP, et al. Breast cancer cell lines contain functional cancer stem cellswith metastatic capacity and a distinct molecular signature. CancerRes 2009;69:1302–13.

34. Tanei T, Morimoto K, Shimazu K, Kim SJ, Tanji Y, Taguchi T, et al.Association of breast cancer stem cells identified by aldehyde dehy-drogenase 1 expression with resistance to sequential paclitaxel andepirubicin-based chemotherapy for breast cancers. Clin Cancer Res2009;15:4234–41.

35. Le Moguen K, Lincet H, Marcelo P, Lemoisson E, Heutte N, Duval M,et al. A proteomic kinetic analysis of IGROV1 ovarian carcinoma cellline response to cisplatin treatment. Proteomics 2007;7:4090–101.

36. Ludwig A, Dietel M, Lage H. Identification of differentially expressedgenes in classical and atypical multidrug-resistant gastric carcinomacells. Anticancer Res 2002;22:3213–21.

37. Russo JE, Hilton J. Characterization of cytosolic aldehyde dehydro-genase from cyclophosphamide resistant L1210 cells. Cancer Res1988;48:2963–8.

38. Yoshida A, Dave V, Han H, Scanlon KJ. Enhanced transcription of thecytosolic ALDH gene in cyclophosphamide resistant human carci-noma cells. Adv Exp Med Biol 1993;328:63–72.

39. Kohn FR, Landkamer GJ, Manthey CL, Ramsay NK, Sladek NE. Effectof aldehyde dehydrogenase inhibitors on the ex vivo sensitivity ofhumanmultipotent and committed hematopoietic progenitor cells andmalignant blood cells to oxazaphosphorines. Cancer Res1987;47:3180–5.

40. Riveros-Rosas H, Julian-Sanchez A, Pina E. Enzymology of ethanoland acetaldehyde metabolism in mammals. Arch Med Res1997;28:453–71.

41. Liu S, Ginestier C, Charafe-Jauffret E, Foco H, Kleer CG, Merajver SD,et al. BRCA1 regulates human mammary stem/progenitor cell fate.Proc Natl Acad Sci U S A 2008;105:1680–5.

42. Yokota A, Takeuchi H, Maeda N, Ohoka Y, Kato C, Song SY, et al.GM-CSF and IL-4 synergistically trigger dendritic cells to acquireretinoic acid-producing capacity. Int Immunol 2009;21:361–77.

43. Chang B, Liu G, Xue F, Rosen DG, Xiao L, Wang X, et al. ALDH1expression correlates with favorable prognosis in ovarian cancers.Mod Pathol 2009;22:817–23.

44. Tanner B, Hengstler JG, Dietrich B, Henrich M, Steinberg P, Weikel W,et al. Glutathione, glutathione S-transferase alpha and pi, and alde-hyde dehydrogenase content in relationship to drug resistance inovarian cancer. Gynecol Oncol 1997;65:54–62.

45. Landen CN Jr, Birrer MJ, Sood AK. Early events in the pathogenesis ofepithelial ovarian cancer. J Clin Oncol 2008;26:995–1005.






2010;9:3186-3199. Published OnlineFirst October 1, 2010.Mol Cancer Ther Charles N. Landen, Jr, Blake Goodman, Ashwini A. Katre, et al. CancerTargeting Aldehyde Dehydrogenase Cancer Stem Cells in Ovarian

Updated version

10.1158/1535-7163.MCT-10-0563doi:

Access the most recent version of this article at:

Material

Supplementary

http://mct.aacrjournals.org/content/suppl/2010/10/06/1535-7163.MCT-10-0563.DC1

Access the most recent supplemental material at:

Cited articles

http://mct.aacrjournals.org/content/9/12/3186.full#ref-list-1

This article cites 43 articles, 17 of which you can access for free at:

Citing articles

http://mct.aacrjournals.org/content/9/12/3186.full#related-urls

This article has been cited by 34 HighWire-hosted articles. Access the articles at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected]

To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://mct.aacrjournals.org/content/9/12/3186To request permission to re-use all or part of this article, use this link



http://mct.aacrjournals.org/lookup/doi/10.1158/1535-7163.MCT-10-0563

http://mct.aacrjournals.org/content/suppl/2010/10/06/1535-7163.MCT-10-0563.DC1

http://mct.aacrjournals.org/content/9/12/3186.full#ref-list-1

http://mct.aacrjournals.org/content/9/12/3186.full#related-urls

http://mct.aacrjournals.org/cgi/alerts

mailto:[email protected]

http://mct.aacrjournals.org/content/9/12/3186
